Handbook
/
Product & Engineering
The YC Advice: Do Things That Don't Scale
Paul Graham's famous essay applies to every early-stage startup. Here's how to embrace unscalable tactics for scalable outcomes.
Paul Graham’s essay “Do Things That Don’t Scale” is required reading for founders. The counterintuitive insight: in the early days, you should do things that would never work at scale. The manual, the personal, the uneconomic—these tactics get you to the point where scale becomes possible.
Why Unscalable Works
You Don’t Have Users Yet
Scalable tactics assume existing distribution: word of mouth, SEO, viral loops. But those require users. When you have no users, you need to acquire them one by one, even if it’s inefficient.
Airbnb founders went door to door in New York, photographing listings and convincing hosts to join. Stripe founders physically installed their payment integration in companies’ codebases. DoorDash founders delivered food themselves.
None of this scales. All of it was necessary.
You Don’t Understand Your Users Yet
Scalable products require deep understanding of users. But you can’t understand users without close contact. Manual, unscalable interactions teach you what automated, scalable products cannot.
When you respond to support tickets personally, you learn what confuses users. When you onboard users manually, you learn what they actually need. When you deliver the service yourself, you understand the value you’re creating.
You Need to Create Delight
Scalable experiences are often generic. Unscalable experiences can be magical. Handwritten thank-you notes, personal calls, surprise gifts—these create the word of mouth that eventually becomes scalable.
Users who feel special tell their friends. Users who feel like numbers don’t.
You Need to Learn What Scales
You can’t automate what you don’t understand. Doing things manually first reveals what should be automated later. The manual process becomes the spec for the automated system.
Types of Unscalable Tactics
Manual Customer Acquisition
Go to your users. If your users are at coffee shops, go to coffee shops. If they’re at conferences, go to conferences. If they’re online, DM them individually.
Stripe’s Collison brothers would literally grab laptops from founders and install Stripe integration on the spot. “Let me just do this for you right now.” No sales pitch, just action.
Reach out personally. Email, DM, call, text. One at a time. Ask to talk. Learn their problems. Convince them to try your product.
Recruit your friends first. Your friends will try things and give honest feedback. Use them. They don’t scale, but they teach you.
Manual Product Delivery
Concierge service. Deliver the value manually before building the automation. If you’re building a meal planning app, plan meals for 10 users by hand. You’ll learn what they need and whether they value it.
Wizard of Oz. Make it look automated while humans do the work behind the scenes. Users experience the value; you learn whether it works before building.
White-glove onboarding. Get on calls with every new user. Walk them through setup. Solve their problems. This doesn’t scale, but it creates activated users who stick around and refer others.
Manual Relationship Building
Customer success before customer success. Before you have a customer success team, be the customer success team. Call your churned users. Email your active users. Build relationships.
Founder support. When users email support, the founder responds. They get faster, better answers. They feel valued. They become advocates.
Community building. Be in every community where your users are. Participate genuinely. Help people. Become known. This is slow but compounds over time.
When to Stop Doing Unscalable Things
Unscalable tactics are means, not ends. Eventually, you need to scale. Signs it’s time to transition:
You’re at capacity. You can’t personally onboard another user without dropping something. Time to automate.
You’ve learned the pattern. You’ve done it enough times that the process is clear. Now you can build a system.
You have product-market fit. Unscalable tactics helped you find fit. Now you need to capture the market.
The unit economics demand it. If your LTV is $100 and manual acquisition costs $200, you need scalable channels.
The transition should be gradual. Don’t stop doing unscalable things; do more scalable things. Keep some unscalable magic even as you grow.
What You Learn From Unscalable
Real Problems
When you talk to users directly, you hear their problems in their words. Not your assumptions, not your survey questions—their actual lived experience. This is the foundation of product-market fit.
Messaging That Works
Through dozens of manual sales conversations, you discover what language resonates. What objections come up. What benefits matter. This becomes your marketing.
Ideal Customer Profile
Manual acquisition shows you who actually buys, who sticks around, who refers others. Your theoretical customer segment evolves into a concrete profile.
Product Requirements
Manual delivery reveals what users actually need, what delights them, what frustrates them. You build the right product instead of the imagined product.
Distribution Channels
By trying many manual approaches, you discover which channels have the best customers. These channels, systematized, become your growth engine.
Resisting the Urge to Scale Prematurely
Founders resist unscalable tactics because:
"It’s inefficient." Yes. That’s the point. Efficiency comes later.
"It won’t work at scale." Correct. You’re not at scale. You’re trying to get there.
"Real companies don’t do this." Real companies that are successful did this when they were small. Google’s founders personally indexed websites. Amazon’s founder packed boxes.
"I should be strategic." Strategy without users is fantasy. Get users first.
"Engineers should be building." If the product isn’t validated, building more product is waste. Validate first, build after.
Balancing Unscalable and Building
You need to do both. But the balance shifts over time:
Pre-PMF (0-100 users): 80% unscalable tactics, 20% building. You’re learning.
Early PMF (100-1000 users): 50/50. Start automating what you’ve learned while maintaining close user relationships.
Growth (1000+ users): 20% unscalable, 80% building and scaling. Keep some direct user contact but focus on systems.
Examples to Copy
Airbnb
Founders photographed listings themselves to improve quality. Traveled to NY to recruit hosts in person. Had breakfast with hosts to understand their needs. Designed the site from users’ apartments.
Stripe
Installed integrations in founders’ code personally. Responded to support tickets within minutes. Built custom features for early users. Hung out at YC dinners to recruit users.
Pinterest
Created content and pins manually to demonstrate the product’s potential. Reached out to bloggers individually. Attended meetups and conferences.
DoorDash
Founders delivered food themselves to understand the driver experience. Went restaurant to restaurant recruiting merchants. Handled customer support personally.
Wufoo
Sent handwritten thank-you notes to every customer. Created personal relationships that drove word of mouth for years.
Key Takeaways
•
Unscalable tactics get you to the point where scalable tactics work
•
Manual acquisition, delivery, and relationship building teach you what to automate
•
Don’t resist inefficiency—embrace it while you’re small
•
The transition from unscalable to scalable should be gradual, not abrupt
•
Even at scale, maintain some unscalable magic—it creates the delight that differentiates you
AIMake has access to all of this
Our AI has access to the entire Startup Handbook. Ask it anything about building your startup.
Get started →
← Previous
Choosing Your Tech Stack as a Non-Technical Founder
Next →
Feature Flags and Progressive Rollouts